The group project that was fixed

Group work is defended as preparation for working life and complained about for a specific reason: the mark is shared and the effort is not. Both positions are right, and the fix is structural.

The standard problem has two parts. First, a single mark for a group gives each member an incentive to rely on the others, and the strongest incentive falls on whoever is least likely to be caught. Second, and less discussed, the student who does the extra work is not only unrewarded but is doing work that teaches them less, because carrying a project usually means doing the parts you already know how to do.

The interventions that work are well tested. Assign roles with separate deliverables, so each person's contribution is identifiable and the parts fit only if everyone produces theirs. Mark a short individual reflection alongside the group product, worth enough to matter. And collect a confidential contribution rating from each member, used not to adjust marks directly but to trigger a conversation where ratings disagree sharply.

What does not work is asking a group to 'divide the work fairly', which transfers an unsolved design problem to a set of fifteen-year-olds with no authority over each other.

The deeper point concerns what the exercise is meant to teach. If the aim is collaboration, then the thing to assess is the collaboration: how the group handled disagreement, how they allocated work, what they did when someone fell behind. A group mark for a finished product assesses the product, which a single able student can usually produce alone, and it is the version of group work that teaches least while looking most like the real thing.